Table 6.

The performance of SVM based models developed using binary profile of atoms and symbols together obtained from terminals of SMILES format.

Feature (parameters)	Main dataset					Validation dataset
	Sen	Spc	Acc	MCC	AUROC	Sen	Spc	Acc	MCC	AUROC
N50 (g = 0.005, c = 6, j = 2)	75.50	75.76	75.62	0.51	0.80	63.21	92.63	77.11	0.58	0.89
N100 (g = 0.01, c = 2, j = 3)	81.26	80.39	80.84	0.62	0.88	77.62	78.79	78.18	0.56	0.89
N200 (g = 0.01, c = 1, j = 2)	85.28	81.57	83.32	0.67	0.92	81.06	77.48	79.15	0.58	0.90
C50 (g = 0.01, c = 5, j = 2)	72.47	72.13	72.30	0.45	0.79	78.10	71.43	74.88	0.50	0.84
C100 (g = 0.01, c = 3, j = 1)	77.93	75.83	76.94	0.54	0.83	84.42	78.72	81.69	0.63	0.89
C200 (g = 0.005, c = 5, j = 1)	80.80	79.66	80.20	0.60	0.89	83.09	82.05	82.53	0.65	0.92
N50C50 (g = 0.005, c = 8, j = 3)	86.45	84.19	85.36	0.71	0.91	83.97	87.84	85.86	0.72	0.92
N100C100 (g = 0.01, c = 2, j = 1)	90.38	86.25	88.35	0.77	0.96	86.90	84.94	85.93	0.72	0.94
N200C200 (g = 0.005, c = 1, j = 2)	91.59	87.46	89.35	0.79	0.96	89.29	82.93	85.86	0.72	0.94

Sen, Sensitivity; Spc, Specificity; Acc, Accuracy; MCC, Matthew’s Correlation Coefficient; AUROC, Area Under the Receiver Operating Characteristic curve; N50/N100/N200, first 50/100/200 elements from N-terminal; C50/C100/C200, first 50/100/200 elements from C-terminal; N50C50/N100C100/N200C200, first 50/100/200 elements from N-terminal as well as from C-terminal joined together.